[LU-7815] pinger thread evict the export between MDTs. Created: 25/Feb/16 Updated: 09/Sep/16 Resolved: 16/Apr/16 |
|
| Status: | Resolved |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | Lustre 2.8.0 |
| Fix Version/s: | Lustre 2.9.0 |
| Type: | Bug | Priority: | Critical |
| Reporter: | Di Wang | Assignee: | Di Wang |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | dne2, soak | ||
| Issue Links: |
|
||||
| Severity: | 3 | ||||
| Rank (Obsolete): | 9223372036854775807 | ||||
| Description |
|
In DNE failover tests, we found MDT is being evicted by another MDT because of pinger timeout. Lustre: soaked-MDT0007: haven't heard from client soaked-MDT0004-mdtlov_UUID (at 192.168.1.110@o2ib10) in 230 seconds. I think it's dead, and I am evicting it. exp ffff88080c87d800, cur 1456388330 expire 1456388180 last 1456388100 Lustre: soaked-MDT0006: haven't heard from client soaked-MDT0004-mdtlov_UUID (at 192.168.1.110@o2ib10) in 232 seconds. I think it's dead, and I am evicting it. exp ffff88078f6f3c00, cur 1456388331 expire 1456388181 last 1456388099 which actually should not happen. It looks like we should not add MDS-MDS export to obd_chain_timed list. I will cook a patch soon. |
| Comments |
| Comment by Gerrit Updater [ 25/Feb/16 ] |
|
wangdi (di.wang@intel.com) uploaded a new patch: http://review.whamcloud.com/18663 |
| Comment by Oleg Drokin [ 25/Feb/16 ] |
|
Please submit the patch against master. |
| Comment by Joseph Gmitter (Inactive) [ 25/Feb/16 ] |
|
Assigning to Di....a patch is in flight. |
| Comment by Gerrit Updater [ 25/Feb/16 ] |
|
wangdi (di.wang@intel.com) uploaded a new patch: http://review.whamcloud.com/18676 |
| Comment by Gerrit Updater [ 02/Mar/16 ] |
|
Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/18676/ |
| Comment by Joseph Gmitter (Inactive) [ 16/Apr/16 ] |
|
The patch has landed to master for 2.9.0 |